| # Sources |
|
|
| RPMem data is prepared from the following sources through normalization, |
| segmentation, probe generation, and split construction. Original attribution |
| and source-specific terms apply; the bundle has no single shared license. |
|
|
| | Source | Terms | |
| | --- | --- | |
| | [UltraChat](https://huggingface.co/datasets/HuggingFaceH4/ultrachat_200k) | MIT | |
| | [TaskMaster2](https://github.com/google-research-datasets/Taskmaster/tree/master/TM-2-2020) | CC-BY-4.0 | |
| | [Schema-Guided Dialogue](https://github.com/google-research-datasets/dstc8-schema-guided-dialogue) | CC-BY-SA-4.0 | |
| | [Toucan](https://huggingface.co/datasets/Agent-Ark/Toucan-1.5M) | Apache-2.0 | |
| | [SWE-Zero](https://huggingface.co/datasets/nvidia/SWE-Zero-openhands-trajectories) | CC-BY-4.0; underlying code licenses also apply | |
| | [D2L synthetic QA](https://huggingface.co/datasets/SakanaAI/self_gen_qa_d2l) | Redistribution confirmed; contact source authors for further use terms | |
| | [PERMA](https://huggingface.co/datasets/ustclsc/PERMA) | Apache-2.0 | |
| | [PersonaMem-v2](https://huggingface.co/datasets/bowen-upenn/PersonaMem-v2) | CC-BY-4.0 | |
| | [PrefEval](https://github.com/amazon-science/PrefEval) | CC-BY-NC-4.0 (noncommercial) | |
|
|
| Source revisions and processing settings are provided in the |
| [source recipe](https://github.com/Quark-Medical/rpmem/blob/main/experiments/phase1/formal_sources_v1.yaml) |
| and benchmark metadata. |
|
|
